Publication detail

Testing Internet applications and services using PlanetLab

KOMOSNÝ, D. MRDOVIC, S. IĽKO, P. GREJTÁK, M. POSPÍCHAL, O.

Original Title

Testing Internet applications and services using PlanetLab

English Title

Testing Internet applications and services using PlanetLab

Type

journal article in Web of Science

Language

en

Original Abstract

In this paper we evaluate a commonly used testing tool of Internet applications and services - PlanetLab. PlanetLab is a large-scale network of Linux servers used for Internet development and research. The geographical diversity of the servers allows engineers and researchers to test networking applications in the real Internet worldwide. In the paper we evaluate PlanetLab from different points of view we identified based on its long-term use. We specifically deal with the problematic aspects, such as PlanetLab service availability, networking performance, geographical accuracy, and DNS addressing correctness. Some of the PlanetLab ‘negative’ properties evaluated in this paper are required and needed as PlanetLab was established to reflect the real (erroneous) communication in the Internet. Nevertheless, one should be aware of them to set-up experiments correctly, run them smoothly, and not to misinterpret the results. Based on the observations presented, we give a set of simple conclusions that should help engineers to achieve valid Internet experiment results.

English abstract

In this paper we evaluate a commonly used testing tool of Internet applications and services - PlanetLab. PlanetLab is a large-scale network of Linux servers used for Internet development and research. The geographical diversity of the servers allows engineers and researchers to test networking applications in the real Internet worldwide. In the paper we evaluate PlanetLab from different points of view we identified based on its long-term use. We specifically deal with the problematic aspects, such as PlanetLab service availability, networking performance, geographical accuracy, and DNS addressing correctness. Some of the PlanetLab ‘negative’ properties evaluated in this paper are required and needed as PlanetLab was established to reflect the real (erroneous) communication in the Internet. Nevertheless, one should be aware of them to set-up experiments correctly, run them smoothly, and not to misinterpret the results. Based on the observations presented, we give a set of simple conclusions that should help engineers to achieve valid Internet experiment results.

Keywords

Internet; Measurement; PlanetLab; IP address; Domain name; Ssh; Performance; Location

Released

09.03.2017

Publisher

Elsevier B.V.

ISBN

0920-5489

Periodical

Computer Standards & Interfaces

Year of study

53

Number

1

State

NL

Pages from

33

Pages to

38

Pages count

6

URL

http://www.sciencedirect.com/science/article/pii/S0920548917300703

Documents

BibTex 


@article{BUT134082,
  author="Dan {Komosný} and Sasa {Mrdovic} and Pavol {Iľko} and Matej {Grejták} and Ondřej {Pospíchal}",
  title="Testing Internet applications and services using PlanetLab",
  annote="In this paper we evaluate a commonly used testing tool of Internet applications and services - PlanetLab. PlanetLab is a large-scale network of Linux servers used for Internet development and research. The geographical diversity of the servers allows engineers and researchers to test networking applications in the real Internet worldwide. In the paper we evaluate PlanetLab from different points of view we identified based on its long-term use. We specifically deal with the problematic aspects, such as PlanetLab service availability, networking performance, geographical accuracy, and DNS addressing correctness. Some of the PlanetLab ‘negative’ properties evaluated in this paper are required and needed as PlanetLab was established to reflect the real (erroneous) communication in the Internet. Nevertheless, one should be aware of them to set-up experiments correctly, run them smoothly, and not to misinterpret the results. Based on the observations presented, we give a set of simple conclusions that should help engineers to achieve valid Internet experiment results.",
  address="Elsevier B.V.",
  chapter="134082",
  doi="10.1016/j.csi.2017.02.006",
  howpublished="online",
  institution="Elsevier B.V.",
  number="1",
  volume="53",
  year="2017",
  month="march",
  pages="33--38",
  publisher="Elsevier B.V.",
  type="journal article in Web of Science"
}